How ingredients influence furan and aroma generation in sponge cake Mayela Cepeda-Vázquez, Barbara Rega, Nicolas Descharles, Valérie Camel
To cite this version:
Mayela Cepeda-Vázquez, Barbara Rega, Nicolas Descharles, Valérie Camel. How ingredients influence furan and aroma generation in sponge cake. Food Chemistry, Elsevier, 2018, 245, pp.1025 - 1033. 10.1016/j.foodchem.2017.11.069. hal-01868629
HAL Id: hal-01868629 https://hal-agroparistech.archives-ouvertes.fr/hal-01868629 Submitted on 5 Oct 2018
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. How ingredients influence furan and aroma generation in sponge cake
AUTHORS
Mayela Cepeda-Vázquez, Barbara Rega, Nicolas Descharles, Valérie Camel*
UMR Ingénierie Procédés Aliments, AgroParisTech, Inra, Université Paris-Saclay, 91300
Massy, France
* Corresponding author. AgroParisTech, 16 rue Claude Bernard, F-75231, Paris Cedex 05,
France
Tel. +33 (0) 1 44 08 17 25 email addresses: [email protected]; [email protected]; [email protected]; [email protected] *
ABSTRACT
A wide range of compounds can be formed during thermal processing of food, some of which are relevant for aroma (e.g. furfural), while others are of great health-concern (e.g. furan). This paper presents the study of formulation as affecting the simultaneous generation of furan and furfural, along with other aroma quality markers, in sponge cake by means of headspace trap / GC-MS. Ingredients were screened according to their category
(fat, salt, sugar, egg-based). Glucose-containing formulation resulted in the highest content of furan and furfural (12.5±0.5 ng g-1 and 9.2±0.2 μg g-1, dry basis, respectively), while their lowest amount was found in the egg-white recipe (3.1±0.1 ng g-1 for furan and 0.287±0.078
µg g-1 for furfural, dry basis). The latter also related negatively to all studied compounds.
This work will be useful for developing novel strategies to deliver safe foods with appealing organoleptic attributes.
KEYWORDS formulation, furanic compounds, glucose, egg white, heat-processed foods, quantification, headspace trap, multivariate analysis
Highlights
• Food safety and aroma markers are studied simultaneously
• Assays are performed on a real food matrix (sponge cake)
• Glucose and egg are the most impacting ingredients on furan and furfural generation
• Fat removal, unsaturation or absence of antioxidants do not affect furan content
• Furfural is highly correlated to furan in all studied formulations
1 INTRODUCTION
Heat-treated foods, such as baked goods, are widely consumed among population, and their investigation is especially important from a food quality and safety point of view.
During thermal processing food constituents may induce chemical changes through reaction pathways (such as caramelization, Maillard reaction and lipid oxidation), leading to the formation of a great number of compounds, some of which are of health concern, whereas others play a role in aroma and color. A current challenge is thus, the development of safe yet appealing food for which food reactivity must be taken into account.
In recent years, much attention has been paid to furan, a heat-generated compound of health-concern due to its possible carcinogenicity, occurrence and consumption by population (International Agency for Research on Cancer, 1995; Joint FAO/WHO Expert
Committee on Food Additives, 2011). It is formed through reaction pathways that are common to other compounds having a role in sensory quality (Belitz, Grosch, & Schieberle,
2009; Crews & Castle, 2007; Kroh, 1994; Martins, Jongen, & van Boekel, 2001; Perez
Locas & Yaylayan, 2004; Stadler, 2012), such as furfural, a naturally occurring odorant in many food items (Burdock, 2010; Stofberg & Grundschober, 1987). Getting insight into their simultaneous formation and their concentration in food is therefore important. Furan has been extensively quantified in a broad range of products, including baked goods (Crews &
Castle, 2007; Joint FAO/WHO Expert Committee on Food Additives, 2011; Zoller, Sager, &
Reinhard, 2007). Furfural has also been widely investigated in food, including qualitative, semi-quantitative and ultimately quantitative studies in bakery products (Ait Ameur, Rega,
Giampaoli, Trystram, & Birlouez-Aragon, 2008; Maire, Rega, Cuvelier, Soto, & Giampaoli,
2013; C. Petisca, Henriques, Pérez-Palacios, Pinho, & Ferreira, 2014; Pico, Bernal, &
Gómez, 2015; Poinot et al., 2008; Pozo-Bayón, Ruíz-Rodríguez, Pernin, & Cayot, 2007;
Rega, Guerard, Delarue, Maire, & Giampaoli, 2009). Up to date, however, furan and furfural simultaneous quantitative determination in food are scarce (Chaichi, Ghasemzadeh-
Mohammadi, Hashemi, & Mohammadi, 2015; Pérez-Palacios, Petisca, Henriques, &
Ferreira, 2013; Catarina Petisca, Pérez-Palacios, Pinho, & Ferreira, 2014b), and no studies are reported in baked goods.
Very recently, we optimized headspace trap (HS trap) extraction for furan and furfural simultaneous quantification in sponge cake by gas chromatography coupled to mass spectrometry (GC-MS), which proved to be sensitive, precise and competitive in comparison to other techniques (Cepeda-Vázquez, Blumenthal, Camel, & Rega, 2017).
This method has been successfully applied to sponge cakes having different properties (fat content and humidity) and can thus be used for formulation studies in similar matrices.
Sponge cake is indeed an interesting bakery product to study the relationship between quality and reactivity, since it is made-up with several ingredients (i.e. sugar, salt, egg, fat and flour), leading to a complex chemical composition, and contains a number of compounds that may also have a role in aroma quality (Maire et al., 2013; Pozo-Bayón et al., 2007; Rega et al., 2009). Moreover, studies in food are all-important because of the diverse nature of ingredients, their possible synergistic effect and impact on overall quality.
This paper presents the study of ingredients in sponge cake as influencing the generation of furan, furfural and other relevant aroma compounds (such as aldehydes and pyrazines) by means of HS trap / GC-MS.
2 MATERIALS AND METHODS
2.1 Chemicals and standards
Furan (purity: 99%), furfural (purity: 99%), 2-methylbutanal (purity: 95%), 2,5- dimethylpyrazine (purity: 99.9%), benzaldehyde (purity: 99%), hexanal (purity: 98%), 2- pentylfuran (purity: 97%), 1-hydroxy-2-propanone (purity: 90%) and d4-furan (purity: 98%) were supplied by Sigma-Aldrich. 3-Methylbutanal (purity: 98%) was obtained from Merck,
2,3,5-trimethylpyrazine was supplied by Acros (purity: 99%) and d4-furfural (purity: 99.7%) was obtained from CIL Cluzeau, CDN Isotopes (Pointe-Claire, Canada). Methanol (purity:
99.9%) was purchased from Carlo Erba (Val-de-Reuil, France). Ultrapure water (> 5
MΩ.cm) was produced using a Millipore Elix 3 System from Millipore SAS (Molsheim,
France). Heptane and tetrahydrofuran were provided by Carlo Erba (Val de Reuil, France) and standard (±)-α-tocopherol by Sigma (Steinhein, Germany).
2.2 Ingredients
Flour was kindly provided by Grands Moulins de Paris (Ivry-sur-Seine, France). Caster sucrose was obtained from Tereos (Lille, France) and glucose (dextrose monohydrate) was kindly provided by Roquette Frères (Lestrem, France). Non hydrogenated palm oil was obtained from Palva Celys (Rezé, France) and sunflower oil by Lesieur (Asnières-sur-Seine,
France). Pasteurized whole egg and egg white were supplied by Agrodoubs (Flagey,
France) and Ovoteam (Massy, France). NaCl was supplied by Salinor (Clichy, France).
Sucrose, glucose and salt were stored at room temperature, whole egg and sunflower oil were kept at 4°C and flour, egg white and palm oil were stored at -20°C. Flour and egg white were thawed at 4°C and palm oil in a water bath at 50°C prior to use.
A batch of purified sunflower oil was also obtained by removing tocopherols (naturally occurring antioxidants). Purification was carried out in three successive vacuum filtrations
(150 mbar and twice 120 mbar) of 200 g of sunflower oil through 70 g of dry alumina
(adapted from Yoshida, Kondo, & Kajimoto (1992)). This purified sunflower oil was put into amber glass bottles filled with nitrogen and stored at 4°C until use.
Absence of tocopherols in purified sunflower oil was confirmed by high performance liquid chromatography (HPLC), for which 1 g of oil was diluted in heptane for a final concentration of 0.1 g mL-1. Sample was vortexed for 60 s at maximum speed and put into an ultrasound bath for 30 s for proper homogenization, followed by filtration through a 0.22 nm nylon filter membrane (Interchim, Montluçon, France). Analyses were carried out under isocratic conditions in an Alliance Waters 2695 system equipped with a photodiode array detector
(298 nm) and a Lichrospher 100 DIOL column (5 μm, 250 x 4 mm, AIT France, Houilles,
France). Mobile phase was a 96.15/3.85 (v/v) heptane/tetrahydrofuran mixture (1 mL min-1) and standard tocopherol concentration was 100 mg L-1 (adapted from (AFNOR, 1996)).
2.3 Sponge cake preparation
2.3.1 Formulations for the screening of ingredients
Reference sponge cake (f1) was formulated according to Cepeda-Vázquez et al. (2017).
Five other formulations (f2-f6) were initially prepared in order to perform the screening of ingredients in relation to their category (fat, salt, sugar, egg-based) and their ability to act as precursors of different reaction pathways, namely lipid oxidation, caramelization, Maillard reaction, Strecker's degradation and other chemical interactions (see Table 1).
Concerning the fat category, three types of formulas were designed based on polyunsaturated fatty acids (PUFAs) content (mainly linolenic and linoleic acids, well-known furan precursors) (Crews & Castle, 2007; Stadler, 2012), and thus their likeliness towards enhancing lipid oxidation. Palm oil (f1) is mainly composed of saturated and monounsaturated fat with around 9% of linoleic acid and a minor fraction of linolenic acid.
Sunflower oil (f3) is particularly rich in linoleic acid (> 65%) (U.S. Department of Agriculture
(USDA), 2016). Since these two oils naturally contain tocopherols (i.e. endogenous antioxidants), an additional formulation (f7) was prepared with purified sunflower oil to further test the lipid oxidation pathway in the absence of antioxidants.
2.3.2 Batter preparation
Batter of reference sponge cake (500 g) was prepared according to Cepeda-Vázquez et al.
(2017) by mixing and beating whole eggs, sucrose and salt using a kitchen appliance
(Kenwood Chef, KM300, UK) for 10 min at maximum speed. Flour was gently added during
1.5 min at minimum speed and the whole was further beaten for 0.5 min. Palm oil was gently added during 15 s at minimum speed and the batter beaten for 1 min at the same speed. Additional sponge cakes were prepared by replacing or removing whole egg, sucrose, salt and palm oil from reference formulation according to Table 1. Formulation containing egg white was further battered for 120 s for proper incorporation.
2.3.3 Baking trials
For each formulation, aliquots of 20 g of batter were poured into 21 aluminum molds (8.0 cm x 4.5 cm x 3.5 cm) and baked at 170°C for 25 min in an instrumented oven (Bongard,
Wolfisheim, France) specifically designed to ensure thermal homogeneity (Fehaili, Courel,
Rega, & Giampaoli, 2010).
Immediately after baking, sponge cakes were put into hermetically sealed glass jars and frozen at -20°C. Twenty sponge cakes were used for composite sampling for chemical analyses, and one for color measurements.
Independent baking trials (n=3) were performed to check the reproducibility of sample preparation on the reference formula (f1) as detailed in section 2.8.
2.4 Composite sampling for chemical analysis
For each baking trial, sponge cakes were ground by means of a Grindomix GM200 knife mill, equipped with a stainless steel bowl and titan knifes (Retsch GmbH, Haan, Germany) using the following conditions: five cakes per grinding operation - 3,000 rpm for 10 s, 6,000 rpm for 20 s and 3,000 rpm for up to 50 s. All batches were then mixed together at 2,000 rpm for 5 s and the resulting composite sample was stored at -20°C until analysis. The same procedure was adopted for all formulas.
2.5 Physical and physicochemical properties
Sponge cake formulations (f1 to f6) were characterized by measuring the properties described below.
2.5.1 Dry matter determination
Moisture content was determined gravimetrically on 1 g of composite samples (n=3) after oven drying at 105°C (Memmert, Schwabach, Germany) to nearest milligram (AE 200, d:
0.1mg, Mettler, Switzerland) (adapted from (Fehaili et al., 2010)).
2.5.2 pH measurement
Aliquots of 1 g of ground samples were mixed with 25 mL of ultrapure water (n=3), stirred for 1 min and further stirred during pH measurement of the resulting mixture (PH3, Hach
Lange, Barcelona, Spain) (adapted from (Ait Ameur et al., 2008)).
2.5.3 Color determination
Color measurements were performed on the upper face of a whole cake at 4 points using a
6834 spectro-guide sphere gloss colorimeter (BYK-Gardner, Germany) and expressed in the CIELab color space (adapted from (Poinot et al., 2008)).
2.6 Selection of quality markers
Furan and furfural were selected as main furanic quality markers for food safety and aroma, respectively. Other typical odorant compounds found in baked goods were selected for being representative of different chemical classes (furans, aldehydes, pyrazines, etc.) and reaction pathways (mainly caramelization, Maillard reaction, lipid oxidation and Strecker's degradation) (Table 2).
2.7 HS trap / GC-MS procedure
Furan and furfural were quantified by stable isotope dilution assays according to Cepeda-
Vázquez et al. (2017), and concentrations were estimated. The remaining aroma compounds were studied using semi-quantitative analysis based on chromatographic peak areas and relative abundances were assessed.
Quantitative and semi-quantitative analyses (n=3) were performed simultaneously by means of a TurboMatrix Headspace Sampler HS 40 Trap (Perkin Elmer, Llantrisant, UK) equipped with an air monitoring trap containing a two-sorbent bed (Carbotrap and
Carbosieve SIII, Perkin Elmer), and a Trace GC Ultra gas chromatography system coupled to an ISQ single quadrupole mass spectrometer (Thermo Scientific, Rodano, Italy).
2.7.1 Standard solutions for furan and furfural quantification
Stock and standard solutions were prepared in a room at 18°C, while HS samples were prepared in a separate room to avoid any cross-contamination from the surrounding air.
Stock solutions of d4-furan, d4-furfural, furan and furfural in methanol were prepared separately at a concentration of 2.5 g L-1, put into hermetic amber glass bottles and stored at -20°C for up to one month. A standard solution of d4-furan at a concentration of 25 ng μL-
1 was prepared daily by diluting the stock solution in ultrapure water. A mixed standard solution containing 0.5 and 15 ng μL-1 of d4-furan and d4-furfural, respectively, was prepared daily by diluting the standard solution of d4-furan and the stock solution of d4- furfural in ultrapure water and kept at 4°C until use. All glassware was previously baked out at 55°C for at least 24 h to avoid any contamination prior to analysis (Cepeda-Vázquez et al., 2017).
2.7.2 HS sample preparation
For vial preparation, sponge cake composite samples were kept in an ice bath to avoid volatile loss. Aliquots of ground sponge cake (0.588 g, dry basis) and ultrapure water (9.412 g) were weighed into 20 mL vials for a 16 water / sample amount ratio (dry basis) and 10 g of total amount (water + sample amount, dry basis) (Cepeda-Vázquez et al., 2017). All amounts were exactly measured in mass. A volume of 40 μL of the d4-furan + d4-furfural standard solution was added for a final set of 20 ng of d4-furan and 600 ng of d4-furfural per vial, respectively. Vials were capped with aluminum polytetrafluoroethylene coated silicone septa at each step of preparation and sealed immediately after labeled standards addition to avoid loss of volatiles. Vials were then vortexed for 5 s at maximum speed. All
HS vials were previously baked out in a muffle furnace at 350°C for at least 1 h to avoid any
HS glassware contamination before analysis (Cepeda-Vázquez et al., 2017).
2.7.3 HS trap conditions
Conditions were set as follows, according to the multivariate optimization of HS trap extraction for furan and furfural simultaneous quantitative analysis (Cepeda-Vázquez et al.,
2017): thermostatting temperature (65°C) and time (15 min), number of pressurization cycles (4) and dry purge time (0.9 min). For all assays, shaker mode was on during vial thermostatting (automatically varying through a frequency range). Needle temperature was set to 75°C and vials were pressurized to 40 psi during 0.5 min, followed by a decay time of
2 min per pressurization cycle. After HS extraction, the trap was heated from 28 to 250°C at a 40°C s-1 rate and a pressure of 23 psi, in order to desorb and release the analytes into the transfer line for 0.2 min. Trap hold time was set to 15 min between GC runs and a sample blank (empty vial) was run every three injections to confirm complete desorption. Vial and water blanks were run daily to check for the absence of contaminants.
2.7.4 GC-MS analysis
Analysis conditions were adapted from Cepeda-Vázquez et al. (2017). A 100% polyethylene glycol column (VF-WAXms) was used for separation (60 m x 0.25 mm x 0.5
μm; Agilent, Netherlands). Helium was used as carrier gas at a constant flow rate of 1.2 mL min-1. Injection was done in split mode (split ratio of 1:8). The GC oven was programmed as follows: initial temperature 40°C (held for 4 min), then raised at 10°C min-1 until 150°C and then raised at 20°C min-1 until reaching a final temperature of 240°C (held for 8 min). MS transfer line and ion source temperatures were set to 250°C and 200°C, respectively. The ionization mode was electron impact (EI), 70 eV. Data acquisition was done in nine segments, including a full scan mode from 25 to 150 m/z at 0.2 s per scan and the corresponding selected ion monitoring (SIM) mode with a dwell time of 0.1 s each, except
for furfural and d4-furfural 0.05 s each (Table 2). Chromatographic peak areas for each compound were calculated by extracting the quantifier ions from the SIM mode acquisition data using Quan Browser, Xcalibur 2.1.0 SP1, build 1160 (Thermo Fisher Scientific Inc.,
USA). Native molecules (i.e. generated in sponge cake during baking) were confirmed by means of the Wiley 8 and NIST 08 mass spectra libraries, calculation of normal alkane retention index (RI) and comparison to NIST Chemistry WebBook SRD 69 (2017) indices, and analysis of pure standards.
2.8 Sample preparation variability
In order to assess variability during batter preparation, baking and grinding, independent assays were run for reference formulation (three independent baking trials). Sponge cakes were sampled, frozen and ground as previously described to obtain a composite sample per baking trial. Each composite sample was then analyzed by HS trap / GC - MS in triplicate, both for furan and furfural quantitation, as well as aroma markers semi- quantitation as detailed in section 2.7.
2.9 Data treatment
One-way ANOVA, Tukey-Kramer's honest significant difference (HSD) and Dunnett's tests were performed on dry matter, pH, color measurements as well as furan and furfural quantitative results. Reference formulation (f1) was defined as the control (Dunnett's test).
For the markers here considered, statistical differences among formulations were assessed by means of ANOVA and Tukey-Kramer's HSD test on raw chromatographic peak areas of quantifier ions, while principal component analysis (PCA) was carried out on the mean- centered areas. All analyses were performed using JMP 10.0.0 (SAS Institute Inc.).
3 RESULTS AND DISCUSSION
3.1 Variability in sample preparation
Low variability was observed for furan and furfural quantitation on single sponge cake preparation experiments (≤ 3.5% for both compounds, n=3), as previously reported for the method (Cepeda-Vázquez et al., 2017). Experiment reproducibility assessed through three independent trials, including batter preparation, baking, grinding and HS trap analysis, both for furan and furfural quantitation was also low: 6.4 and 6.8%, respectively. Higher variability was observed for semi-quantitative analyses: 2-methylbutanal and 3- methylbutanal (10%), hexanal (6%), 2-pentylfuran (11%), 1-hydroxy-2-propanone (11%),
2,5-dimethylpyrazine and 2,6-dimethylpyrazine (15%), 2,3,5-trimethylpyrazine (15%), benzaldehyde (8%). Yet, moderate variability was observed overall, showing good reproducibility throughout experiments.
3.2 Sample physical and physicochemical characterization
Low moisture content, neutral pH and mid-browning level (L*) were observed in general
(Table 3). However, according to the ANOVA and Tukey-Kramer's test, significant differences were found among sponge cake formulations within each physical and physicochemical variable and were also grouped differently across them. Surprisingly, major differences were only found for glucose (f5) and egg white formulation (f6), exhibiting maximum and minimum values for pH, L* and b*. Indeed, glucose recipe showed the highest browning level (L*< 50) and the lowest pH (acid), as opposed to egg white formulation which exhibited the lowest browning level (L*> 50) and the highest pH (alkaline).
Low pH in formulation with glucose might be due to the formation of acidic compounds and correlated to advanced browning, as previously reported in cookies (Ait Ameur et al., 2008).
Indeed, while a release of H+ occurs during caramelization (Kroh, 1994), organic acids are also formed during the Maillard reaction (Martins et al., 2001), which may lead to pH fall.
Egg white's pH (7.6 to 7.9), which rises with temperature (Belitz et al., 2009), could explain higher pH in formulation f6. Low dry matter content in this ingredient (12.1%) (Belitz et al.,
2009) may also explain low dry matter in this formulation, and in turn result in a lower browning level, given its higher water content.
3.3 Furan and furfural content in sponge cake
Furan content in sponge cake remains in the range of 3.1-12.5 ng g-1, dry basis (Figure 1) which can be considered as a low to mid-level when compared to different types of bread (≤
30 ng g-1), and similar or low when compared to other products, such as baby food containing fruit (≤ 16 ng g-1), cereal-based snacks (≤ 143 ng g-1) or coffee brew (≤ 199 ng g-
1) (Zoller et al., 2007). Furfural content in sponge cake (0.29-0.83 μg g-1, dry basis), except in the case of glucose recipe (9.2 μg g-1, dry basis), is similar to levels observed for cake/pastry (0.26-2.6 μg g-1) (Petisca et al., 2014a).
3.4 Furan and aroma generation in sponge cake
A scheme based on current literature on the main reaction pathways (i.e. caramelization,
Maillard reaction, Strecker's degradation and lipid oxidation) leading to the selected compounds (Table 2) and considering main precursors in sponge cake ingredients is here put forward (Figure 2). It will be discussed in the following sections to better understand how formulation impacts the generation of key quality markers in sponge cake.
3.4.1 Furan and furfural generation
Furfural can only be generated through caramelization or Maillard reaction (Belitz et al.,
2009; Kroh, 1994; Martins et al., 2001; Srivastava et al., 2017; Whitfield, 1992; Yaylayan &
Keyhani, 2000), while furan can be additionally formed by PUFAs oxidation, as well as amino acid thermal degradation (Crews & Castle, 2007; Perez Locas & Yaylayan, 2004;
Srivastava et al., 2017; Stadler, 2012) (Figure 2). Yet, our results show that furfural generation in sponge cake seems to be favored over furan's, since furfural content (ppb to ppm) was found to be 65 to 730-fold that of furan (ppb), depending on the formulation under study. Both furan and furfural content were mainly impacted by two categories of ingredients, namely sugar and egg-based (Figure 1).
The glucose-containing recipe (f5) exhibited their highest concentration, suggesting that the extent of non-enzymatic browning reactions (caramelization, Maillard reaction) in this formulation is greater than in the reference cake (f1) containing sucrose. This result is not surprising since glucose (reducing sugar) can directly undergo enolization and further degradation (through caramelization or Maillard reaction) (Figure 2) (Kroh, 1994; Martins et al., 2001), while sucrose (non-reducing sugar) first has to undergo hydrolysis before releasing its reducing monomers (glucose and fructose) (Quintas, Guimarães, Baylina,
Brandão, & Silva, 2007). Similar results were previously reported in sponge cake where sucrose replacement by reducing sugars led to higher amounts of furanic compounds (i.e. furfural and HMF (5-hydroxymethylfurfural)) (Zhang et al., 2012). Our results also show a higher molar yield regarding furan generation in the case of the glucose-containing formulation (f5) as compared to the sucrose recipe (f1) (18.0 and 13.0 ng g-1 mol-1, dry basis, respectively). Opposite results were reported in model solutions (Perez Locas &
Yaylayan, 2004), suggesting that sucrose undergoes hydrolysis less effectively in sponge cake than in simple models. Indeed, sucrose degradation into reducing sugars (fructose
and glucose) may be limited in sponge cake (Zhang et al., 2012). This can be easily explained: model systems (single or binary mixtures with a high water content, heated at elevated temperatures, i.e. 250°C) do not reproduce what really occurs during baking of sponge cake (complex chemical composition and structure, where gradual water evaporation and drying of the surface lead to crust formation, whose temperature tends to that of the oven, i.e. 170°C) (Fehaili et al., 2010). This underlines the importance of performing studies in complex matrices.
Interestingly, the increase in furfural content in our recipe (f5) with respect to our reference cake (f1) was much greater than that in furan (around 17-fold and 3-fold, respectively). This suggests that, although glucose is a precursor of both compounds, the pathway leading to furfural (i.e. 1,2-enolization) is favored over the one majorly contributing to furan (i.e. 2,3- enolization) (Belitz et al., 2009; Kroh, 1994; Martins et al., 2001; Perez Locas & Yaylayan,
2004). Indeed, in a recent kinetic study in cake models containing either glucose-only or glucose and leucine, furfural generation was also favored over furan's (Srivastava et al.,
2017).
The egg white formulation (f6) showed the lowest furan and furfural concentrations. Indeed, whole egg is richer in free amino acids (U.S. Department of Agriculture (USDA), 2016), which might enhance the Maillard reaction pathway or might even form furan through amino acid thermal degradation (Perez Locas & Yaylayan, 2004) (Figure 2). Besides, while
PUFAs are present in egg yolk, phospholipids also are and might exert a protective effect by forming antioxidant species when reacting with carbonyl compounds (Maire et al., 2013), hindering lipid oxidation as a result.
Interestingly, all formulations (f1-f3) designed to study furan formation via lipid oxidation had low and very similar furan contents, suggesting that furan formation through PUFA oxidation is not particularly favored. Likewise, in other products such as baby food model
systems containing starch, sugars and soybean oil, a higher furan content was only obtained when oxidizing oil in extreme conditions (14 days at 60°C) prior to use (Owczarek-
Fendor et al., 2012). Also, naturally occurring tocopherols in our fat ingredients (sunflower and palm oil) could exert an antioxidant effect as reported by Maire et al. (2013), hence preventing further furan generation through PUFAs oxidation. However, this hypothesis is here discarded since formulations with raw (f3) or purified sunflower oil (f7) resulted in no differences in furan content (7.1 ± 0.3 and 7.1 ± 0.2 ng g-1, dry basis, respectively, P:
0.337), even if (f7) did show a higher amount of other lipid oxidation markers (i.e. hexanal and 2-pentylfuran).
Salt removal from recipe (f4) did not seem to exert any effect on furan formation (P: 0.0779), while it did on furfural, since a lower amount was found with respect to (f1) (P: 0.0005).
Previous studies on salt effect are scarce and results rather unclear. Undetectable furan levels were reported both in liquid models containing glucose or sucrose added with NaCl
(Fan, Huang, & Sokorai, 2008). In the case of furfural, salt addition showed different effects depending on the complexity of the matrix: no effect on furfural generation in a glucose- containing solution, and different effects on furfural content in cookies (increasing in the presence of leavening agents and decreasing in their absence) (Kocadağlı & Gökmen,
2016). Indeed, the complexity of the food matrix is to be considered (e.g. use of leavening agents).
To conclude, furan generation in sponge cake would seem to proceed through common pathways to furfural (i.e. caramelization and Maillard reaction). Their generation is enhanced in the presence of glucose (reducing sugar) and hindered in the absence of egg yolk (probably due to differences in Maillard reaction progression, given the restriction on free amino acids).
3.4.2 Impact of formulation on selected quality markers
PCA was carried out on all considered markers in order to determine correlations among them and relate them to changes in formulation. According to this analysis, components 1,
2 and 3 account for 95.5% of variability (Figure 3). Three highly correlated groups of compounds are observed in the PC1 vs PC2 loading plot: 1) furan, furfural and hexanal, 2)
Strecker aldehydes, 2-pentylfuran and 1-hydroxypropanone, and 3) pyrazines. However, in the PC1 vs PC3 loading plot, hexanal and 1-hydroxypropanone exhibit a particular behavior: hexanal is opposite to furan and furfural, and 1-hydroxypropanone to Strecker aldehydes, when considering PC3.
These results, along with possible reaction pathways (Figure 2), give rise to several remarks. First of all, furan strongly correlates to furfural and overall behaves differently from all the other volatile markers. 2-Pentylfuran and hexanal are both well-known lipid oxidation products (Belitz et al., 2009; Cerny, 2010; Perez Locas & Yaylayan, 2004) generated from linoleic and arachidonic acid via 2,4-decadienal (Grein, Huffer, Scheller, & Schreier, 1993;
Whitfield, 1992); they strongly correlate when considering PC2, but not regarding PC1. This suggests that additional pathways could be active in the formation of 2-pentylfuran in sponge cake, either through cyclization of other alkenals (Adams, Bouckaert, Lancker,
Meulenaer, & Kimpe, 2011; Hidalgo, Gallardo, & Zamora, 2005) or via a lipid oxidation and amino group interaction (Adams et al., 2011; Whitfield, 1992). It should also be noted that hexanal may be formed prior to sponge cake baking, that is during the beating of batter, as previously shown by Maire et al. (2013).
Pyrazines are produced through a well-known pathway involving aminoketone degradation
(Cerny, 2010; Whitfield, 1992). This very specific origin is well represented on the PCA loading plots where they are closely correlated to one another (Figure 2). Benzaldehyde would be produced either by Strecker's degradation or lipid oxidation, as previously
reported by Whitfield (1992), explaining its high correlation both to other Strecker aldehydes and 2-pentylfuran.
1-Hydroxy-2-propanone is an especially interesting marker. It could be produced either through sugar degradation (Yaylayan & Keyhani, 2000) or Maillard reaction (Belitz et al.,
2009; Martins et al., 2001) via 1-deoxy-2,3-hexodiulose, a common intermediate to furan, but also from aminoketones produced during Strecker's pathway (Whitfield, 1992).
Moreover, as shown in simple models, 1-hydroxy-2-pronanone could yield pyrazines in the presence of NH3 along with aliphatic aldehydes (e.g. hexanal) and these compounds may be formed during cake preparation (through Strecker degradation from cysteine and lipid oxidation, respectively (Whitfield, 1992)). All of the above would explain 1-hydroxy-2- propanone's particular location in the PCA loading plots (Figure 3), correlated to Strecker aldehydes, positively influencing PC1 (like pyrazines, unlike furan) and PC2 (like furan, unlike pyrazines).
The effect of sponge cake formulation is clearly visible in the PC1 vs PC2 plane (85%).
Indeed, four clusters of recipes are observed in the score plot: glucose (f5, top left), egg white (f6, bottom left), with no added fat nor salt (f2, f4, center) and reference and sunflower oil (f1, f3, right center). This indicates that in the presence of glucose, the peak intensities of furan, furfural and hexanal are higher, and those of the remaining compounds are lower.
However, given hexanal's particular behavior in the PC1 vs PC3 plane, it can only be concluded that the presence of glucose might induce a quicker progression of caramelization and Maillard reaction in sponge cake, yielding higher amounts of furan and furfural and limiting the formation of compounds related to amino acid degradation (mainly
Strecker-related and pyrazines). In the absence of egg yolk (f6), the peak intensities of nearly all compounds are lower. In this case, the restriction on some precursors (e.g. amino
acids) might be responsible for limiting the progression of certain pathways, in particular those related to amino acid degradation. Indeed, our work confirms previous findings where a higher response of 2-methylbutanal, 3-methylbutanal, benzaldehyde and 2-pentylfuran were observed in a whole egg sponge cake recipe, compared to an egg white-only formulation (Maire et al., 2013).
All considered markers were significantly influenced by formulation as shown by the
ANOVA procedure, confirming PCA results (see supplementary table). Overall, sponge cake reference formulation containing sucrose and whole egg has a greater variety in aroma compounds such as Strecker aldehydes and pyrazines; glucose recipe increases furan and furfural response, most likely due to this sugar's reducing nature; and egg white formulation causes a decrease in the peak intensity of aroma compounds, probably related to amino acid-related pathways.
Clearly, an appropriate selection of key markers, a suitable experimental design and the use of multivariate analysis tools make it possible to have better understanding of the link between formulation and reaction pathways in real food, as in this case with sponge cake.
4 CONCLUSIONS
Furan and furfural content, as well as the generation of key aroma markers from different reaction pathways taking place during baking, are clearly influenced by sponge cake formulation. Sugar and egg-based are undoubtedly the most interesting categories of ingredients involved in their formation in this product.
Furan content is mainly affected by the presence of glucose (increase) and the absence of egg yolk (decrease). Reaction pathways active in its formation in sponge cake are most likely caramelization, Maillard reaction or even amino acid degradation, but not lipid
oxidation. Furfural is highly correlated to furan, but its higher content in all recipes suggests that its generation is favored over furan's.
Key aroma compounds (Strecker aldehydes, pyrazines and 2-pentylfuran) are positively related to formulation containing whole egg, probably due to a higher contribution of free amino acids in egg yolk.
This work represents a step forward in delivering baked goods having optimal food safety and aroma properties. It contributes to the understanding of the two-way relationship among key quality markers and formulation. Our findings will be useful for developing furan mitigation strategies based on formulation and also for understanding how changes in recipes (e.g. sucrose replacement, low fat, fat free or salt free recipes) might affect the generation of key compounds in these bakery products.
FUNDING
This work was supported by the Mexican National Council for Science and Technology
(Consejo Nacional de Ciencia y Tecnología - CONACYT) and the French Embassy in
Mexico (joint grant: 314446).
ACKNOWLEDGEMENTS
The authors wish to thank Paola Soto for technical support in oil purification and running
HPLC analysis, and also David Blumenthal and Pierre Giampaoli for fruitful discussions.
The authors have declared no conflict of interest.
REFERENCES
Adams, A., Bouckaert, C., Lancker, F. Van, Meulenaer, B. De, & Kimpe, N. De. (2011).
Amino Acid Catalysis of 2-Alkylfuran Formation from Lipid Oxidation-Derived α,β-
Unsaturated Aldehydes. Journal of Agricultural and Food Chemistry, 59, 11058–11062.
http://doi.org/10.1021/jf202448v
AFNOR. (1996). Corps gras d’origines animale et végétale - Détermination des teneurs en
tocophérols et en tocotriénols par chromatographie en phase liquide à haute
performance, NF EN ISO 9936 Octobre 2006.
Ait Ameur, L., Rega, B., Giampaoli, P., Trystram, G., & Birlouez-Aragon, I. (2008). The fate
of furfurals and other volatile markers during the baking process of a model cookie.
Food Chemistry, 111(3), 758–763. http://doi.org/10.1016/j.foodchem.2007.12.062
Belitz, H.-D., Grosch, W., & Schieberle, P. (2009). Food Chemistry (4th ed.). Springer-
Verlag.
Burdock, G. (2010). Fenaroli’s handbook of flavor ingredients (6th ed.). Boca Raton: CRC
Press.
Cepeda-Vázquez, M., Blumenthal, D., Camel, V., & Rega, B. (2017). Multivariate
optimization of headspace trap for furan and furfural simultaneous determination in
sponge cake. Talanta, 164, 708–715. http://doi.org/10.1016/j.talanta.2016.10.073
Cerny, C. (2010). Thermal Generation of Aroma-Active Volatiles in Food. In A. Herrmann
(Ed.), The Chemistry and Biology of Volatiles (pp. 231–252). John Wiley & Sons Ltd.
Chaichi, M., Ghasemzadeh-Mohammadi, V., Hashemi, M., & Mohammadi, A. (2015).
Furanic compounds and furfural in different coffee products by headspace liquid-phase
micro- extraction followed by gas chromatography – mass spectrometry : survey and
effect of brewing procedures. Food Additives & Contaminants: Part B, 8(1), 73–80.
http://doi.org/10.1080/19393210.2014.981601
Crews, C., & Castle, L. (2007). A review of the occurrence, formation and analysis of furan
in heat-processed foods. Trends in Food Science & Technology, 18(7), 365–372.
http://doi.org/10.1016/j.tifs.2007.03.006
Fan, X., Huang, L., & Sokorai, K. J. B. (2008). Factors affecting thermally induced furan
formation. Journal of Agricultural and Food Chemistry, 56(20), 9490–9494.
http://doi.org/10.1021/jf801612c
Fehaili, S., Courel, M., Rega, B., & Giampaoli, P. (2010). An instrumented oven for the
monitoring of thermal reactions during the baking of sponge cake. Journal of Food
Engineering, 101(3), 253–263. http://doi.org/10.1016/j.jfoodeng.2010.07.003
Flavor & Extract Manufacturers Association (FEMA). (2017). FEMA Flavor Ingredient
Library. Retrieved from http://www.femaflavor.org/flavor (last accessed 03/07/2017)
Grein, B., Huffer, M., Scheller, G., & Schreier, P. (1993). 4-Hydroxy-2- alkenals and other
products formed by water-mediated oxidative decomposition of α,β-unsaturated
aldehydes. Journal of Agricultural and Food Chemistry, 41, 2385–2390.
Hidalgo, F. J., Gallardo, E., & Zamora, R. (2005). Strecker Type Degradation of
Phenylalanine by 4-Hydroxy-2-nonenal in Model Systems. Journal of Agricultural and
Food Chemistry, 53, 10254–10259. http://doi.org/10.1021/jf052240+
International Agency for Research on Cancer. (1995). Dry cleaning, some chlorinated
solvents and other industrial chemicals. In IARC Monographs on the evaluation of
carcinogenic risks to humans (Vol. 63, pp. 393–407).
Joint FAO/WHO Expert Committee on Food Additives. (2011). Safety evaluation of certain
contaminants in food. In WHO Food Additives Series: 63, FAO JECFA Monographs 8.
Seventy-second meeting.
Kocadağlı, T., & Gökmen, V. (2016). Effects of Sodium Chloride, Potassium Chloride, and
Calcium Chloride on the Formation of α-Dicarbonyl Compounds and Furfurals and the
Development of Browning in Cookies during Baking. Journal of Agricultural and Food
Chemistry, 64(7838–7848). http://doi.org/10.1021/acs.jafc.6b03870
Kroh, L. W. (1994). Caramelisation in food and beverages. Food Chemistry, 51, 373–379.
Maire, M., Rega, B., Cuvelier, M.-E., Soto, P., & Giampaoli, P. (2013). Lipid oxidation in
baked products : Impact of formula and process on the generation of volatile
compounds. Food Chemistry, 141, 3510–3518.
http://doi.org/10.1016/j.foodchem.2013.06.039
Martins, S. I. F. S., Jongen, W. M. F., & van Boekel, M. A. J. S. (2001). A review of Maillard
reaction in food and implications to kinetic modelling. Trends in Food Science &
Technology, 11, 364–373.
National Institute of Standards and Technology (NIST). (2017). NIST Chemistry WebBook.
Retrieved from http://webbook.nist.gov/chemistry/ (last accessed 02/27/2017)
Owczarek-Fendor, A., De Meulenaer, B., Scholl, G., Adams, A., Van Lancker, F., Eppe, G.,
… De Kimpe, N. (2012). Furan formation in starch-based model systems containing
carbohydrates in combination with proteins, ascorbic acid and lipids. Food Chemistry,
133(3), 816–821. http://doi.org/10.1016/j.foodchem.2012.01.098
Pérez-Palacios, T., Petisca, C., Henriques, R., & Ferreira, I. M. P. L. V. O. (2013). Impact of
cooking and handling conditions on furanic compounds in breaded fish products. Food
and Chemical Toxicology, 55, 222–228. http://doi.org/10.1016/j.fct.2012.12.058
Perez Locas, C., & Yaylayan, V. A. (2004). Origin and mechanistic pathways of formation of
the parent furan - a food toxicant. Journal of Agricultural and Food Chemistry, 52,
6830–6836.
Petisca, C., Henriques, A. R., Pérez-Palacios, T., Pinho, O., & Ferreira, I. M. P. L. V. O.
(2014). Assessment of hydroxymethylfurfural and furfural in commercial bakery
products. Journal of Food Composition and Analysis, 33, 20–25.
http://doi.org/10.1016/j.jfca.2013.10.004
Petisca, C., Pérez-Palacios, T., Pinho, O., & Ferreira, I. M. P. L. V. O. (2014). Optimization
and Application of a HS-SPME-GC-MS Methodology for Quantification of Furanic
Compounds in Espresso Coffee. Food Analytical Methods, 7, 81–88.
http://doi.org/10.1007/s12161-013-9601-0
Pico, J., Bernal, J., & Gómez, M. (2015). Wheat bread aroma compounds in crumb and
crust : A review. Food Research International, 75, 200–215.
http://doi.org/10.1016/j.foodres.2015.05.051
Poinot, P., Arvisenet, G., Grua-Priol, J., Colas, D., Fillonneau, C., Le Bail, A., & Prost, C.
(2008). Influence of formulation and process on the aromatic profile and physical
characteristics of bread. Journal of Cereal Science, 48(3), 686–697.
http://doi.org/10.1016/j.jcs.2008.03.002
Pozo-Bayón, M. A., Ruíz-Rodríguez, A., Pernin, K., & Cayot, N. (2007). Influence of Eggs
on the Aroma Composition of a Sponge Cake and on the Aroma Release in Model
Studies on Flavored Sponge Cakes. Journal of Agricultural and Food Chemistry, 55,
1418–1426. http://doi.org/10.1021/jf062203y
Quintas, M., Guimarães, C., Baylina, J., Brandão, T. R. S., & Silva, C. L. M. (2007).
Multiresponse modelling of the caramelisation reaction. Innovative Food Science &
Emerging Technologies, 8(306–315). http://doi.org/10.1016/j.ifset.2007.02.002
Rega, B., Guerard, A., Delarue, J., Maire, M., & Giampaoli, P. (2009). On-line dynamic HS-
SPME for monitoring endogenous aroma compounds released during the baking of a
model cake. Food Chemistry, 112(1), 9–17.
http://doi.org/10.1016/j.foodchem.2008.05.028
Srivastava, R., Bousquières, J., Cepeda-Vázquez, M., Roux, S., Bonazzi, C., & Rega, B.
(2017). Kinetic study of furan and furfural generation during baking of cake models.
Food Chemistry, (in press). http://doi.org/10.1016/j.foodchem.2017.06.126
Stadler, R. H. (2012). Heat-generated toxicants in foods: acrylamide, MCPD esters and
furan. In D. Schrenk (Ed.), Chemical contaminants and residues in food (pp. 201–232).
Woodhead Publishing Limited.
Stofberg, J., & Grundschober, F. (1987). Comsumption ratio and food predominance of
flavoring materials. Perfumer & Flavorist, 12(4), 27.
U.S. Department of Agriculture (USDA). (2016). National Nutrient Database for Standard
Reference. Retrieved from https://ndb.nal.usda.gov/ndb/ last accessed 03/21/2017)
Whitfield, F. B. (1992). Volatiles from interactions of Maillard reactions and lipids. Critical
Reviews in Food Science and Nutrition, 31(1–2), 1–58.
http://doi.org/10.1080/10408399209527560
Yaylayan, V. A., & Keyhani, A. (2000). Origin of Carbohydrate Degradation Products in L -
Alanine/D-[13C] Glucose Model Systems. Journal of Agricultural and Food Chemistry,
48, 2415–2419. http://doi.org/10.1021/jf000004n
Yoshida, H., Kondo, I., & Kajimoto, G. (1992). Participation of Free Fatty Acids in the
Oxidation of Purified Soybean Oil During Microwave Heating. Journal of the American
Oil Chemists’ Society, 69(11), 1136–1140. http://doi.org/10.1007/BF02541050
Zhang, Y.-Y., Song, Y., Hu, X.-S., Liao, X.-J., Ni, Y.-Y., & Li, Q.-H. (2012). Effects of sugars
in batter formula and baking conditions on 5-hydroxymethylfurfural and furfural
formation in sponge cake models. Food Research International, 49, 439–445.
http://doi.org/10.1016/j.foodres.2012.07.012
Zoller, O., Sager, F., & Reinhard, H. (2007). Furan in food: Headspace method and product
survey. Food Additives and Contaminants, 24, 91–107.
http://doi.org/10.1080/02652030701447389
Table 1 Sponge cake formulation according to ingredient categories
Ingredients (g) Flour Fat Salt Sugar Egg-based Total Formulation description Testing for Type 55 PO SO pSO NaCl S G E EW effect of fat type and amount on lipid reference (palm oil, oxidation; effect of salt and sugar type f1 salt, sucrose, whole on browning reactions; effect of egg- 125 20 0 0 5 125 0 225 0 500 egg) based ingredients on aroma generation f2 no added fat effect of fat amount on lipid oxidation 125 0 0 0 5 125 0 225 0 480 PUFAs reactivity towards lipid f3 sunflower oil (linoleic) oxidation in the presence of 125 0 20 0 5 125 0 225 0 500 tocopherols (antioxidants) f4 no salt salt effect on browning reactions 125 20 0 0 0 125 0 225 0 495 sugar type effect on browning f5 glucose 125 20 0 0 5 0 125 225 0 500 reactions effect of egg-based ingredients on f6 egg white 125 20 0 0 5 125 0 0 225 500 aroma generation purified sunflower oil PUFAs reactivity towards lipid f7 (linoleic, without oxidation in the absence of 125 0 0 20 5 125 0 225 0 500 tocopherols) tocopherols (antioxidants) PO: palm oil; SO: sunflower oil; pSO: purified sunflower oil; S: sucrose; G: glucose; E: whole egg; EW: egg white; PUFAs: polyunsaturated fatty acids
Table 2 Selected volatile markers in sponge cake, associated pathways, sensory attributes and identification method
Selected Compound Class Reaction pathway Flavor/odor CAS No. RI * RI Identification ref. exp. ions (m/z)** furan furan MR, C, LO, AA - 110-00-9 813 p 811 39, 68 MS, RI, STD degradation a, b, c, d
2-methylbutanal aldehyde S e, f, g almond, cocoa, fermented, 96-17-3 930 p 928 41, 44, 57 MS, RI, STD hazelnut, malty n, o
e, f, g o p 3-methylbutanal aldehyde S malty, roasty cucumber-like 590-86-3 934 933 41, 44, 57 MS, RI, STD hexanal aldehyde LO e, f, g apple, fat, fresh, green, oil, 66-25-1 1103 p 1099 44, 56 MS, RI, STD grassy, tallow n, o
2-pentylfuran furan LO, LO & AA butter, floral, fruit, green bean, 3777-69-3 1240 p 1248 81, 138 MS, RI, STD interaction b, e, g, h, i, j mushroom, raw nuts n, o
1-hydroxy-2-propanone hydroxyketone MR, C, S e, g, k, l pungent, sweet caramellic, 116-09-6 1319 p 1345 43, 74 MS, RI, STD ethereal o
2,5-dimethylpyrazine pyrazine S interaction g cocoa, roast beef, roasted nut, 123-32-0 1357 p 1357 42, 108 MS, RI, STD crust-like, popcorn n, o
2,6-dimethylpyrazine pyrazine S interaction g cocoa, coffee, green, roast beef, 108-50-9 1362 p 1362 42, 108 MS, RI roasted nut, roasted n, o
2,3,5-trimethylpyrazine pyrazine S interaction f, g cocoa, earthy, must, potato, 14667-55-1 1429 p 1439 42, 122 MS, RI, STD roast n, o furfural furan MR, C e, g, k, l, m almond, baked potatoes, bread, 98-01-1 1503 p 1507 39, 95, 96 MS, RI, STD burnt, spice, soil, roasted, sweet, toasted n, o benzaldehyde aldehyde S, LO g bitter almond, burnt sugar, 100-52-7 1533 p 1587 77, 106 MS, RI, STD cherry, malt, roasted pepper, caramel n, o
Internal standards d4-furan 6142-90-1 - - 42, 72 - d4-furfural 1219803-80-1 - - 42, 98, 100 - MR: Maillard reaction; C: caramelization; LO: lipid oxidation; AA: amino acid; S: Strecker's degradation; RI: normal alkane retention index; MS: mass spectra; STD: pure standard a Crews & Castle, 2007; b Perez Locas & Yaylayan, 2004; c Srivastava et al., 2017; d Stadler, 2012; e Belitz, Grosch, & Schieberle, 2009; f Cerny, 2010; g Whitfield, 1992; h Adams, Bouckaert, Lancker, Meulenaer, & Kimpe, 2011; i Grein, Huffer, Scheller, & Schreier, 1993; j Hidalgo, Gallardo, & Zamora, 2005; k Martins, Jongen, & van Boekel, 2001; l Yaylayan & Keyhani, 2000; m Kroh, 1994; n Flavor & Extract Manufacturers Association (FEMA), 2017; o Pico, Bernal, & Gómez, 2015; p National Institute of Standards and Technology (NIST), 2017 *Retention indices on a ZB-WAX-like column; **Quantifier ions are in bold
Table 3 Sponge cake physical and physicochemical characterization
dry matter Formulation pH L* a* b* (%) f1 92.8 a 7.0 d 53.2 c 12.8 a 15.5 c f2 90.0 bc 7.1 d 56.3 c 13.7 a 18.6 bc f3 89.5 c 7.2 c 55.5 c 14.7 a 19.0 bc f4 93.6 a 7.7 b 62.6 b 12.9 a 21.7 b f5 91.3 b 5.5 e 43.9 d 6.2 b 4.6 d f6 87.3 d 8.0 a 70.1 a 12.9 a 26.5 a Values are the means from 3 replicates for dry matter and pH, and 4 replicates for color measurements. Values followed by different letters within columns are significantly different (P: <0.05)
a a
b b b b
/g sample)/g c
ng
furan contentfuran (
a /gsample) c b c μg d
( d furfuralcontent
b furfural content (μg/g sample)
b c c d d
Figure 1 Furan (ng/g) and furfural (µg/g) content in sponge cake according to formulation. (a) furan and furfural content in formulations 1 to 6; (b) formulations having the lowest furfural content; f1: reference; f2: no added fat; f3: sunflower oil; f4: no salt; f5: glucose; f6: egg white; values are the means and 95% confidence intervals (n=3); mean values with different letters are significantly different (P: <0.05)
Figure 2 Reaction pathways leading to key quality markers in sponge cake. PUFA: polyunsaturated fatty acids; 2PF: 2-pentylfuran; 2MB: 2- methylbutanal; 3MB: 3-methylbutanal; 1H2P: 1-hydroxy-2-propanone; 2,5 DMP: 2,5-dimethylpyrazine; 2,6 DMP: 2,6-dimethylpyrazine; 2,3,5 TMP: 2,3,5-trimethylpyrazine (Crews & Castle, 2007; Perez Locas & Yaylayan, 2004; Srivastava et al., 2017; Stadler, 2012; Belitz, Grosch, & Schieberle, 2009; Cerny, 2010; Whitfield, 1992; Adams, Bouckaert, Lancker, Meulenaer, & Kimpe, 2011; Grein, Huffer, Scheller, & Schreier, 1993; Hidalgo, Gallardo, & Zamora, 2005; Martins, Jongen, & van Boekel, 2001; Yaylayan & Keyhani, 2000; Kroh, 1994)
Figure 3 Principal component analysis on selected volatile compounds in sponge cake formulations. f1: reference; f2: no added fat; f3: sunflower oil; f4: no salt; f5: glucose; f6: egg white; 2MB: 2-methylbutanal; 3MB: 3-methylbutanal; 2PF: 2-pentylfuran; 1H2P: 1-hydroxy-2-propanone; 2,5 DMP: 2,5-dimethylpyrazine; 2,6 DMP: 2,6- dimethylpyrazine; 2,3,5 TMP: 2,3,5-trimethylpyrazine
Supplementary Table Peak areas of the selected markers in sponge cake
Formulation furan 2MB+3MB hexanal 2PF 1H2P 2,5 DMP+2,6 DMP 2,3,5 TMP furfural benzaldehyde 1 1019293 b 117334872 a 1036261 d 6123387 b 793669 a 2165852 a 401325 a 983668 bc 1737559 a 2 897436 cd 80511591 c 3833174 c 2383420 d 621995 bcd 1791143 b 280398 bc 1332845 b 1364217 c 3 969202 bc 91631525 b 1073709 d 7135143 a 746201 ab 1924555 b 308601 b 927058 bc 1516750 bc 4 869030 d 95836968 b 4831596 b 6144870 b 476103 d 1458376 c 245609 c 598459 c 1632867 ab 5 2291855 a 110936634 a 5566375 a 4429624 c 641014 abc 35616 e 21244 e 16721826 a 1029226 d 6 506805 e 9527479 d 426097 e 737079 e 488099 cd 1146420 d 98061 d 468726 c 384190 e 2MB: 2-methylbutanal; 3MB: 3-methylbutanal; 2PF: 2-pentylfuran; 1H2P: 1-hydroxy-2-propanone; 2,5 DMP: 2,5-dimethylpyrazine; 2,6 DMP: 2,6-dimethylpyrazine; 2,3,5 TMP: 2,3,5-trimethylpyrazine
Values are the means of raw chromatographic peak areas from 3 replicates. Values followed by different letters within columns are significantly different (P<0.05)